Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Will AI steal submarines’ stealth?

spectrum.ieee.org

Post reply on HN
Print view View original

51–60 of 159 posts

Re: Will AI steal submarines’ stealth?

#51

Post by jlawson »

Retric wrote: post #10

At least for the US why not keep one or more nuclear missile subs in the Great Lakes? Lake Superior alone covers 31,700 square miles to an average depth of 483 feet. It’s far to large an area to target with a first strike and well protected from non US/Canadian forces. Edit: A US/Canadian treaty demilitarized the Great Lakes. https://en.wikipedia.org/wiki/Rush–Bagot_Treaty

They might be an easier target than you think.

What could a dozen 1MT warheads, sunk to a depth of 100ft, do to submarines anywhere in the Great Lakes? Shockwaves travel through water much better than through air. Water explodes into steam in a way that air doesn't.

My intuition is that you actually could pretty reliably paste a sub inside the Great Lakes using some nukes launches from Russia. The lakes may just not be deep or big enough to hide in.

This is leaving aside the treaty, the lack of freshwater sub technology, the long time to retaliation due to long range, the lack of a sub facility on the lakes, and the unnecessary nature of the whole enterprise (if the US lost 1 boomer sub to an enemy they could bring the others into a defensive posture making them practically impossible to hunt down, to preserve 2nd strike capability).

Re: Will AI steal submarines’ stealth?

#52

Post by sgc »

ceejayoz wrote: post #33
jwithington wrote: post #21

Earlier quoted context omitted.

I think we could but why build a whole base complete with nuclear weapons handling and reactor maintenance facilities on the Great Lakes? Just use the ones in Georgia/Washington. It’s a fun thought exercise though!

The idea is that ones based in the Great Lakes would be immune to interception by enemy submarines discovering and shadowing them; you're not gonna get an Akula up the St. Lawrence Seaway.

However, it would leave them is a small, well-known area and would allow for easier signature profiling via sensors mounted on any number of craft, that could then be used on other subs of the same class elsewhere in the world. That was the reason why when some idiot in the White House revealed where our subs were near China and HK, it was a big deal.

Re: Will AI steal submarines’ stealth?

#53

Post by sgc »

unchocked wrote: post #8

Seems naive that, as the article claims, nations might agree to limit their detection technology to enhance survivability of their adversaries’ nuclear forces. If “ocean transparency” comes to pass, it seems much likelier that we’d have to deal with a realignment of the MAD calculus.

There could be no trust for such an agreement, hence it will never happen.

Re: Will AI steal submarines’ stealth?

#54

Post by trashtester »

lumost wrote: post #50

If submarines became non stealth we would likely see carrier groups and other missile ships providing high speed resilient second strike capability. A potentially destabilizing effect might be to motivate the creation of second/first strike satellite hosted weapons.

If you cannot hide a sub, you certainly cannot hide a carrier group, and carrier groups are probably not much easier to protect against attacks either, unless their air defences are radically improved (by lasers, rail guns etc to protect vs incoming missiles).

Re: Will AI steal submarines’ stealth?

#55

Post by bpodgursky »

BeefWellington wrote: post #48
unchocked wrote: post #8

Seems naive that, as the article claims, nations might agree to limit their detection technology to enhance survivability of their adversaries’ nuclear forces. If “ocean transparency” comes to pass, it seems much likelier that we’d have to deal with a realignment of the MAD calculus.

I would guess that on some level, some analyst many years ago laid out a plan for what should happen with that, and the first thing would be to have a few armed satellites to ensure the continued effectiveness of MAD as a concept. Yes, it would violate treaties against the weaponization of space but I would lay even money on the idea that some country has already done this.

It's tricky because if you do it in secret, it's not MAD, and it's not actually an effective deterrent.

It only works as a deterrent if you break the treaty and everyone knows you've broken the treaty (whether you publicly admit it or not) but we aren't there afaik.

Re: Will AI steal submarines’ stealth?

#56

Post by teaearlgraycold »

oldjavacoder wrote: post #42

Modern AI can barely drive a car under ideal conditions without crashing. The ocean is full of sounds. AI is decades away (still, and I've said this now for decades) from figuring out what it is hearing.

If you read the article you’d have seen that exact idea mentioned at the top. This concern is 30+ years in the future.

Re: Will AI steal submarines’ stealth?

#57

Post by foolfoolz »

unchocked wrote: post #8

Seems naive that, as the article claims, nations might agree to limit their detection technology to enhance survivability of their adversaries’ nuclear forces. If “ocean transparency” comes to pass, it seems much likelier that we’d have to deal with a realignment of the MAD calculus.

MAD has been in constant re-alignment since it became a thing. first with detection to be able to retaliate before getting blown up. then with guaranteed second strike from subs. then with missile defense which we don’t have yet but in theory if you can shoot down the nukes you can strike without retaliation

Re: Will AI steal submarines’ stealth?

#59

Post by whakim »

ceejayoz wrote: post #34
whakim wrote: post #27

Earlier quoted context omitted.

Even though this seems paradoxical, there's actually a logic to it. One of the reasons countries like the United States and Russia (previously, the USSR) maintain such large nuclear arsenals is to project a credible thread of a second strike - i.e., the idea that any first strike wouldn't be able to completely knock out the opponent's nuclear arsenal. If both sides think that the "submarine" portion of their nuclear…

It would very much change the calculus for the UK and France, though. Maybe Israel, as well.

I'm not sure it'd change the calculus for the UK or France because they are NATO members which effectively ensures their second strike capability. (It certainly might change the cost/benefit calculation on building submarines, though!) Israel probably doesn't even subscribe to a second strike doctrine given that the potential adversaries they're trying to deter don't even have nuclear weapons.

Re: Will AI steal submarines’ stealth?

#60

Post by toast0 »

bpodgursky wrote: post #55
BeefWellington wrote: post #48

Earlier quoted context omitted.

I would guess that on some level, some analyst many years ago laid out a plan for what should happen with that, and the first thing would be to have a few armed satellites to ensure the continued effectiveness of MAD as a concept. Yes, it would violate treaties against the weaponization of space but I would lay even money on the idea that some country has already done this.

It's tricky because if you do it in secret, it's not MAD, and it's not actually an effective deterrent. It only works as a deterrent if you break the treaty and everyone knows you've broken the treaty (whether you publicly admit it or not) but we aren't there afaik.

> It only works as a deterrent if you break the treaty and everyone knows you've broken the treaty (whether you publicly admit it or not) but we aren't there afaik.

It could work if it gets to the point where 'only an idiot' wouldn't have broken the treaty, so you can just assume everyone has, because the idea is fairly obvious, the technology is easy, and there's not any other good options. Or if a rogue state does it or something like it, then you can assume the US and Russia have done it or will be doing it shortly.

Post reply on HN